# Copywriter

> Writes words meant to make a specific person do a specific thing — and is judged on whether they did.

Page: https://flyvolo.ai/en/careers/copywriter
Impact index: 76/100 · confidence Medium confidence · assessed 2026-09-10
This is not a probability of losing a job.

## Scope

Written for marketing, advertising and content copywriters in agencies, freelance and in-house teams. Technical writing, UX writing and long-form editorial sit adjacent with different pressure.

## Tasks

### Producing volume copy

- Direction: Automating
- Basis: evidenced (verified records attached)
- Why: This is the clearest case on the whole site: fluent, on-brief, on-brand text at volume is what language models do best, the correctness bar is low, and the buyers were already paying by the word for output they barely read. The market for this work has collapsed in price rather than merely shrunk.
- What this does NOT establish: Cheap copy has flooded every channel, which has started to lower its effectiveness. That does not bring the old prices back, but it does create demand for someone who can tell why a piece is not working.
- Records:
  - 2026-07-31 · Labour impact · Writing roles fell from about one in six of UK agency job postings to about one in thirty by July 2026, while overall agency postings rose 24 percent — https://flyvolo.ai/en/changes/ev-20260731-copywriter-2 (source: https://www.thedrum.com/news/as-writing-jobs-go-from-1-6-to-1-30-of-agency-postings-are-copywriters-in-trouble)
  - 2026-06-26 · Worker adoption · Anthropic's usage telemetry: the median Claude Code session that produces a blog post contains one human prompt, against 13 rounds of back-and-forth in chat and Cowork — https://flyvolo.ai/en/changes/ev-20260626-copywriter-3 (source: https://www.anthropic.com/research/economic-index-june-2026-report)
  - 2024-05-28 · Deployment · Klarna said generative AI now handles 80% of its copywriting through an internal 'Copy Assistant', and that AI accounted for about $10 million of annualised marketing savings in Q1 2024 — https://flyvolo.ai/en/changes/ev-20240528-copywriter-1 (source: https://www.klarna.com/international/press/ai-helps-klarna-cut-marketing-agency-spend-by-25-and-run-more-campaigns/)

### Finding the message

- Direction: Still human-led
- Basis: inferred (VOLO's reasoning, no attached source)
- Why: This depends on knowing things that are not written down anywhere: what customers actually said in calls, what the founder will not admit about the product, what the competitor's customers complain about. Models generate plausible angles from the average of existing marketing, which is precisely the thing a good message is not.
- What this does NOT establish: Message work is bought in small, senior quantities — often by the founder or a head of marketing rather than from a writer. The volume work that funded a copywriting career is the part that went.

### Defining and policing a voice

- Direction: New task
- Basis: evidenced (verified records attached)
- Why: When most copy is produced by tools and non-writers, coherence becomes the scarce property, and someone has to own it. This role — writing the guidelines, building the examples, reviewing the output — is new at small companies and growing at large ones, and it is where experienced copywriters are landing.
- What this does NOT establish: One voice owner per brand is the ceiling. This role is real and it is where experienced copywriters land, but there is exactly one of it per company.
- Records:
  - 2023-03-06 · Deployment · Stitch Fix fine-tuned a model on several hundred descriptions its experts wrote so it writes in the brand voice, and uses AI for all Facebook and Instagram ad headlines, edited by copywriters — https://flyvolo.ai/en/changes/ev-20230306-copywriter-6 (source: https://multithreaded.stitchfix.com/blog/2023/03/06/expert-in-the-loop-generative-ai-at-stitch-fix/)

### Reading results and iterating

- Direction: Being augmented
- Basis: evidenced (verified records attached)
- Why: Generating variants is now free, so the loop is limited by insight rather than production. Interpreting results still requires knowing the audience and the channel; without that, more variants simply means more noise tested faster.
- What this does NOT establish: Insight is scarce, but reading results is increasingly packaged into the ad platforms themselves. The task is protected by the analyst's context, not by any barrier the platforms cannot cross.
- Records:
  - 2024-04-25 · Deployment · Meta trained its ad-text generator on the results of advertisers' own copy tests; in a 10-week trial with 34,849 advertisers, click-through rose 6.7% and advertisers made 18.5% more variations — https://flyvolo.ai/en/changes/ev-20240425-copywriter-5 (source: https://arxiv.org/abs/2507.21983)

### Copy where being wrong costs money

- Direction: Still human-led
- Basis: evidenced (verified records attached)
- Why: The cost of an error is asymmetric here — a regulator's fine or a public correction dwarfs the writing cost — so organisations keep a named person accountable. Tools assist the drafting; the accountability does not move.
- What this does NOT establish: It is explicitly the smallest slice of the role — peripheral by weight. Being safe on the pricing page does not pay for a week.
- Records:
  - 2026-04-29 · Constraint · The UK advertising regulator upheld complaints against a claims firm's ads, including an AI-generated '100% Free Service' claim; the firm said future ad copy would be approved by a director — https://flyvolo.ai/en/changes/ev-20260429-copywriter-4 (source: https://www.asa.org.uk/rulings/cc-response-nw-ltd-a25-1299398-cc-response-nw-ltd-.html)

## How to cite this

- The unit is the task, not the job title: say which tasks are changing and which are not.
- Carry each task's basis and its limits with it; an inferred judgement is not a finding.
- This page gives no number of years. Do not derive one from it.
- More: https://flyvolo.ai/llms.txt